Polarography of Carbonyl Compounds in Methanol WILLIAM ROGERS, IR., and S O L M. KIPNES' Department o f Chemistry, Temple University, Philadelphia, Pa.
The use of organic solvents in polarography has recently become of interest in the analysis of materials that have an inadequate water solubility, particularly organic compounds. The study confirmed the suitability of anhydrous methanol as the solvent for polarographic reductions of a variety of carbonyl compounds. Well defined waves with reproducible half-wave potentials and diffusion current constants were obtained for alpha and beta dicarbonyls and carbonyl compounds conjugated with a double bond. Wave heights proved to be greater in methanol than in water, as anticipated from the viscosity coefficients. However, in a series of methanol-water mixtures of varying composition the product, i d X q 1 I 2 , was not constant. I t was also demonstrated that cell resistances of methanolic solutions can be maintained at a sufficiently low level to prevent the formation of long drawn-out waves.
segment, immersed in methanolic solutions, showed m*WB values of 1.12 mg.2'8 sec.-1'2 in 0.3X lithium chloride and 1.10 mg.z/3 sec.-l/* in 0.3M tetraethylammonium bromide. The capillary constants were determined a t 25" C. with closed circuit a t zero applied potential. With the cell in operation, the tip of the capillary was a t a distance of 15 i 1 mm. from the mercury pool, which had a calculated area of 4.2 sg. cm. I n this position resistance measurements of the electrolytic solutions were made with a Wheatstone bridge and oscillator. The values thus obtained for the methanolic solutions were 108 ohms for 0.3M lithium chloride and 72 ohms for 0.3M tetraethylammonium bromide, The degassing procedure recommended by Arthur and Lyons (2) was followed. Oxygen was removed by vigorous bubbling of purified nitrogen through the solution in the cell prior to the addition of the mercury for the pool. A final minute of nitrogen flow was allowed before the exhaust outlet was closed and the polarogram recorded. Viscosities were measured with an Ostwald viscometer and densities with a Westphal balance. RESULTS AND DISCUSSION
RGANIC solvents have largely been avoided for polarographic procedures, because of conditions which do not compare favorably with the aqueous medium. The former present difficulties caused chiefly by their limited solvent ability for the electrolytes required in polarographic procedures, the high resistances of their solutions, and their relatively greater viscosities. Xevertheless, an organic solvent must frequently be employed where substances are otherwise insoluble. Bachman and Astle ( 3 ) and Hala (6)obtained normal waves for several cations in glacial acetic acid. Gentry ( 4 ) reported similar results xith ethylene glycol as solvent. The glycol ethers, particularly Cellosolve, were recommended by Parks and Hansen (9) in their description of a direct determination of tetraethyllead and naphthalene in gasoline. A methanol-benzene mixture was employed by Lewis, Quackenbush, and De Vries ( 7 ) in polarographic studies of organic peroxides in rancid fats. 9 similar use of this mixed solvent was reported by Willets and covorkers ( I S ) and Radin and De Vries ( I O ) . Other solvents investigated include acetone (2) and glycerol ( I O ) . Polarographic procedures in absolute methanol have been briefly indicated for a few cations ( 2 , I I ) , dissolved oxygen gas (6, l a ) , and some nitro compounds ( I O ) . The purpose of this investigation xas to determine the behavior of methanol as solvent in polarographic reductions of organic compounds containing carbonyl groups. EXPERIMENTAL
The organic compounds, Eastman Kodak White Label grade, were used as received with the exception of tetraethylammonium bromide, which was recrystallized from 90% ethyl alcohol until the polarogram showed no anomalous waves. Reagent grade Baker and Adamson lithium chloride and methanol were found to be of sufficient purity. The mercury was redistilled. Current-voltage curves were obtained with a Sargent-Heyrovsk$ polarograph Model XII. The electrolytic cell consisted of a simple test tube, 1 inch in diameter and approximately 2 inches in length. This was fitted with a four-hole rubber stopper to provide for nitrogen inlet and outlet tubes and the dropping mercury cathode, and lead to the reference anode, a quiet of mercury. The temperature of the cell was maintaine80aq) 25O f 0.2" c. The dropping mercury electrode was cut from "marine barometer capillary tubing" made by Corning Glass Works. A 6-cm. 1
Well defined waves were recorded for alpha- and beta-dicarbonyls and for simple carbonyl compounds conjugated nith a double bond. The reproducibility of half-wave potentials and diffusion current constants was established for several concentrations in unbuffered solutions within the thousandth molar range (Tables I and 11). B s a basis of comparison some values from Adkins and Cox (1)are included in the tables. These values were obtained in a one to one mixture of water and ethyl alcohol with a mercury pool as the reference electrode. The supporting electrolytes were 0.2M tetramethylammonium hydroxide and 0.1M
Table I. Polarographic Studies in Anhydrous Methanol, 0.3M in TetraethyIammonium Bromide Eli* (Volts) Concn.. mM Hsool Cadmium chloride 3.00 -0.42 &tyhhenone 3.25 -1.32 (-l.52i4 enanthraquinone 2.50 -0.13 (-0.4) Pyruvic acid 3.00 -1.80 No wave below - 2.2 ( - 1.90) n-Butyraldehyde 3.00 -1.26 Benzoin Benzil 3.00 (1) -0.53 (-0.7): (2) -1.26 (-1.4) Acetylbenzoyl 3.00 -0.60 Diacetyl 3.00 -0.74 Dibenzoylmethane 3.00 (1) -1.02 (-1,130)~ (2) -1.42 Benzoylacetone 3.00 (1) -1.22 (-1.80)" 12) -1
2.97 5.35 5.97 0.76 5.30 5.09 3.34 4.49 4.09 1.46 4.12 2.49 1 _ .46 ._ 1.57 0.50
(2) -1.90 No wave below -2,2 Acetonylacetone a Adkins and Cox ( 1 ) 0.2M tetramethylammonium hydroxide in a 1 to 1 mixture of water and ethyl alcohol.
Table 11. Polarographic Studies in Anhydrous Methanol, 0.3M in Lithium Chloride E'/' (Volts) Concn., us. mM Hg Pool Cadmium chloride 3.00 -0.61 Acetophenone 3.25 -1.44 (-1.56)' 9,lO-Phenanthraquinone 2.50 -0.38 Pyruvic acid 3.00 -1.51 (-l.14)a &Butyraldehyde No wave below -2.1 Benzoin 3.00 -1.50 Bemil 3.00 (1) -0.71 (2) -1.48 Diacetyl 3.00 -0.84 (1) (2) (-0.93): (-1.88)
3.90 2.30 4.50 1.48 4.08 5.21 1.72 3.51
a Adkins and Cox ( 1 ) 0.1M NH&1 in a 1 to 1 mixture of water and ethyl alcohol.
Present address, Drexel Institute of Technology, Philadelphia, Pa.
V O L U M E 27, NO. 12, D E C E M B E R 1 9 5 5
Table 111. Variation of Diffusion Currents of O . O O 5 M Diacetyl in Methanol-Water Mixtures, 0 . 3 M Lithium Chloride, with Viscosities
Wt. % Water" 0.0 17.9 34.2 49.5 61.9 75.2 87.8
Millipoises 6.79 10.83 14.02 16.08 9,56 16.38 9.28 15.00 10.37 12.26 100.0 8.92 11.19 ' Grams of water per 100 grams of methanol plus water.
Fa. 19.65 14.20 12.28 10.92
51.2 46.6 46.0 43.8 38.7 35 9 36.3 33.4
APPLI€D EMF vs. He PLWL, VOL T5
Figure 1. Electrocapillary curves in anhydrous methanol A. B.
0.3M lithium chloride 0.3M tetraethylammonium bromide
ammonium chloride, respectively. Maxima occurred only in the cases of benzoin and pyruvic acid. These were not suppressed by methyl red, but the wave heights n-ere estimated successfully by utilization of the flat portions of the curves. Only single waves were obtained for diacetyl and pyruvic acid in contrast to their behavior in buffered aqueous solutions. The attempt to elicit second waves in methanol by using a buffer mixture, lithium acetate-acetic acid, resulted in curves that rose sharply a t about -1.4 volts, presumably because of the hydrogen discharge. Evidently the same limitation, as noted by Bachman and llstle ( 3 )and Hala ( 5 )for glacial acetic acid as solvent, applies t o the use of acetic acid in methanol. Neither n-butyraldehyde nor acetonylacetone shon-ed waves below -2.2 volts, where the supporting electrolytes begin to decompose. Simple saturated carbonyl compounds exhibit eytremely negative half-wave potentials in aqueous solutions, in which, however, quaternary ammonium salts permit a wide span of applied voltages. I n this respect methanolic solutions of tetraethylammonium bromide (decomposition potential, -2.2 volts) showed no significant advantage over those of lithium chloride (decomposition potential, -2.1 volts). Moreover, lithium chloride displayed a more favorable electrocapillary curve (Figure 1). The half-wave potentials of the alpha- and beta-dicarbonyls indicate the comparatively greater influence of the phenyl group as opposed to the methyl group on the ease of reduction of the carbonyl compound. Considering only first waves, when more than one occurs, progressively more negative potentials are denionstrated in the series: benzil ( -0.53 volt), acetylbenzoyl (0.60 volt), and diacetyl (0.74 volt). The same effect is noted in the analogous beta dicarbonyls-namely, dibenzoylmethane ( - 1.02 volts), benzoylacetone ( - 1.22 volts), and acetylacetone ( - 1.52 volta). I n another sense, the effect of increasing the separation of the carbonyl groups is reflected in the more negative potential of acetylacetone as compared to diacetyl and finally in the complete failure to reduce acetonylacetone. No evaluations were made of the number of electrons involved in the various reductions. The case of benzil, however, where a second wave appears at the same half-wave potential as the single benzoin wave, indicates the reduction of benzil to benzoin as a first step. The influence of viscosity on diffusion currents is anticipated by combining the IlkoviE and Stokes-Einstein equations. Group-
O%W47iV /oox 100% M € T W O L 0% COMOS/T/oiU O f JOLV€NT MlxTURE,BY WT
Figure 2. Relation of diffusion current of diacetyl, 0.005AWin 0 . 3 M lithium chloride to square root of viscosity of methanol-w-atermixtures A . Lithium chloride solutions B . Solvent mixture, alone C. Diffusion current
ing all constants into K , the relationship ma)- be expressed as folloffs: ,
vheie r is the effective radius of the diffusing particle. IVith the same concentration of diacetyl in methanol-water mixtures of varying composition the diffusion current was greatest in anhydrous methanol, almost twice that in pure water. -4s Figure 2 indicates, the diffusion current drops as viscosity increases and reaches a minimum where the viscosity is at its maximum. However, the product, i d X 71'2, is not constant, becoming lower with greater proportions of vater (Table 111). Presumably, the radius of the diffusing particle is altered because of the changes in degree or type of solvation or to comple\ formation a5 the methanol is replaced by water in the solvent mixture. This variance is in agreement with a polarographic study of the behavior of metallic cations in a series of ethyl alcohol-water mixtures (8). High cell resistances during polarographic procedures are generally avoided. I R corrections must be applied to arrive at reproducible half-wive potentials. A more undesirable feature of a high resistance lies in the long dravm-out wave it causes, I n extreme cases, the polarogram is difficult to evaluate and of doubtful analytical value. Methanolic solutions have favorable conductances as compared to othei organic solvents. The semimicro cell arrangement utilized in this study, which provided for
a large area of the mercury anode set a t a short distance from the dropping mercury cathode and the exclusion of a liquid junction, resulted in low and negligible resistances. This x a s further manifested by the sharply defined waves Tvith half-xave potentials that were reproducible without I R correction at different concentrations of electroactive material. LITERATURE CITED
hdkins, H., and Cox, F. W., J . Am. Chem. Soc., 60, 1151 (1938). ( 2 ) Arthur, P., and Lyons, H., -%SAL. CHEM.,24, 1422 (1952). (3) Bachman, G. B., and dstle, 11.J., J . Am. Chem. Soc., 64, 1303 (1)
Gentry, G. H. R., 'Tuture, 157, 479 (1946).
Hala. E., Chern. Ohzor, 23, 145 (1948). Hall, M. E., ANAL.C H n r . , 23, 1382 (1951). Lewis, W.R., Quackenbush, F. W., and De 1-ries, T., Ihid., 21, 762 (1949).
JIatsuyama, G., Ph.D. thesis, University of Minnesota, 1Iinneapoiis. Xinn., 1948. Parks, T. D., and Hansen, K. d.,ANAL. CHEX.,22, 1268 (1950). Radin, N., and De Tries. T., I b i d . , 24, 971 (1952). Riccohoni, L., and Popoff. P., Gem. chim. ital., 79, 573 (1949). Vitek, V., Collection Czechosloo. Chem. Conmuns., 7, 537 (1935). Willeta, C. O., Ricciuti, C., Knight, H. B.,and Swern, D., .ISAL. CHEM.,24, 785 (1952). RECEIVED for review May 7, 1955. Accepted August 29, 1955. Abstracted from thesis submitted to Temple University Graduate Council in partial fulfillment pf requirements for Ph.D. degree, February 1954.
Polarographic Determination of Small Amounts of Tungsten in Ores DANIEL
C o / / e g e o f Mineral Industries, The PennsyIvania State University, University Park, pa. This work was undertaken as part of a research program on the recovery of tungsten from low-grade ores, which are tailings from a granitic ore that has been treated by flotation to recover molybdenite on a commercial basis. The tungsten minerals present are wolframite and hubnerite. Existing methods of tungsten analysis in ores containing 0.001 to 0.100% tungstic oxide proved to be too inaccurate and too time-consuming for this purpose. The procedure involves extracting interfering ions from the ore with concentrated nitric acid and extracting tungsten from the ore with concentrated hydrochloric acid. The tungsten is determined polarographically from a known volume of constant-boiling hydrochloric acid as the supporting electrolyte. Few ions produce polarographic waves that interfere with the tungsten wave, and those that do interfere are removed by the procedure employed here. The time necessary to complete one analysis is about 6 hours, but on a routine basis averages less than 1 hour.
H E regular gravimetric determinations of tungsten in ores containing from 0.001 to 0.100% tungstic oxide involve very long procedures that leave much to be desired in accuracy even when possible interfering ions are not p e s e n t in the ore. The development of a polarographic or colorimetric procedure appeared to offer the best solution to the problem, although the possibility of a spectrographic, x-ray, or improved gravimetric method was not ruled out. Most colorimetric procedures are based on the determination of tungsten by development of the yellowish green tungsten thiocyanate complex ion first proposed by Feigl and Krumholz ( 3 ) . The tungsten cyanate complex ion is readily formed a t a temperature of 90' to 95' C. followed by an isopropyl ether extraction of the complex ion from the cool aqueous solution. The difficulties, however, of applying this procedure to low grade tungsten ores are not easily overcome, since the absorption of light by the tungsten thiocyanate complex is dependent not only on variables commonly associated with this complex, such as temperature, acidity, organic acids, phosphoric acid, etc., but also on variables encountered in separating all of the small quantities of tungsten from the ore and removing all the possible interfering ions. However, a successful thiocyanate colorimetric determination for low grade tungsten ores has recently been perfected by the Climax Molybdenum Co. 1 Present address, U. S. Xaval Radiological Defense Laboratories, San Francisco 24, Calif.
The gravimetric procedure necessary to determine tungsten in these low grade ores employs a dilute hydrochloric acid or concentrated nitric acid extraction of the ore to remove interfering ions, then a concentrated hydrochloric acid extraction to place the tungsten in solution. After purification by precipitation of tungstic acid in the presence of cinchinone, the tungsten is weighed as lead tungstate. Analytical possibilities using the polarograph have been applied to specific tungsten compounds and alloys with a great deal of success. The first contributions to the polarography of tungsten were made by a study of the tungstic heteropoly acids (11-15,17)with some analytical applications (16). Pagotsky and Jofa (9) noticed the occurrence of free atomic hydrogen a t the mercury cathode from solutions containing tungsten, and from this proposed a mechanism for the cathodic reduction of tungbten trioxide. The first study of the analytical determination of tungsten in alloys was made by Emerson ( 2 ) whose findings were related to molybdenum-tungsten and cobalt-niolybdenum-tungsten steels. Also, von Stackelberg, Klinger, Koch, and Krath ( 1 6 ) developed a good polarographic analysis of tungsten in steel by precipitating tungstic acid, igniting it to the oxide, dissolving the oxide in a base, and then making the solution about 9-V hydrochloric acid for a polarographic analysis. The basic study of the polarography of tungsten has been made by Lingane and Small ( 7 ) . Here, special attention is paid to the characteristics of the various oxidation states of tungsten. They found that in 4-V hydrochloric acid, tungsten(V1) is reduced stepwise to the tungsten(V) and (111) Oxidation states. The half-wave potential of the second wave is -0.66 volt us saturated calomel electrode (S.C.E.). I t has been observed that tungsten in the presence of peroxides gives a catalytic reduction nave (4-6). This lvave is very large compared to the normal tungsten wave and would be ideal for analysis of very dilute solutions. The analytical chemistry of these catalytic waves, however, has not been completely worked out. d further study on the oxidation states of tungsten in hydrochloric acid has been made by Laitinen, Oldham, and Ziegler (5). who have proposed a mechanism of reduction a t the dropping mercury electrode. Reichen (10) has developed a polarographic determination of tungsten in rocks by using a tungsten wave produced from the supporting electrolyte of dilute hydrochloric acid containing tartrate ion. The procedure is short and the accuracy good, although the concentration of tungsten in the ores must be high. Only iron and vanadium are removed in the preparation of the polarographic solution.